Using OWL to define topic maps in RDF
Posted on April 22 2005 by Lars Marius Garshol
in reponse to From RDF to Topic Maps and back again
You write that the survey "is curiously brief about one quite obvious solution: an ontology in the W3C's OWL".
There are two reasons for this, the first being that at the time of the survey's publication no proposals taking this approach had been published. (One not written by me is under publication now, however.)
The other reason is that this approach, unfortunately, is not workable. OWL is an ontology language defined as an RDF vocabulary. In other words, it sits inside RDF itself, and can only be used to define the meanings of RDF identifiers. As you write above, the problem with RDF and topic maps interoperability is that different world-views are built into them. OWL cannot solve this, because it is based on RDF and so already assumes that worldview.
Also, contrary to what is often thought, OWL doesn't actually let you define what anything means. It just lets you express (some of) the constraints on the vocabulary you use (people may like various kinds of fruit) and also the logical consequences of certain statements (if someone likes something the someone must be a sentient being). This is unfortunately mostly orthogonal to RDF/TM interoperability.
This is not to say that OWL is wholly useless for solving this problem, but it can only help in a couple of places (some of them mentioned in the survey).
This was a rather superficial response (for reasons of space :), but if you want more detail there is some to be found in:
Replies to this post: